RegPilot
Overview
TypeSafe System One triages and scores. Azure OpenAI drafts. Humans clear what the confidence gate holds back.
Guardrail
PII + injection
Triage
TypeSafe choices
Draft
Azure memo
Gate
Human or auto
Needs you
—
pending review or rework
Blocked
—
injection / guardrail
Fast path
—
routine high-confidence
Avg confidence
—
0 cleared · 0 total
Grounding holds
—
soft-fail checks (audit)
System One
AI decision layer
Fallback mode
TypeSafe is not configured. Local classifiers and Azure small deployment handle decisions until TYPESAFE_API_KEY is set.
Routing policy
- Critical urgency → full Azure analysis
- Escalate noul ≥ 0.75 on non-routine → full path
- High-conf routine Consumer/Ops → fast path
- Confidence > 0.90 → auto-approve
- Weak grounding → soft-cap confidence ≤ 0.49
Mix
By category
Loading…
Action
Items that need you
Loading…
Recent
Pipeline feed
Loading…